Free-roaming dogs limit habitat use of giant pandas in nature reserves

Page created by Sally Alvarado
 
CONTINUE READING
Free-roaming dogs limit habitat use of giant pandas in nature reserves
www.nature.com/scientificreports

            OPEN            Free-roaming dogs limit habitat use
                            of giant pandas in nature reserves
                            Ramana Callan1, Jacob R. Owens1, Wenlei Bi1,2, Benjamin Kilham3, Xia Yan1, Dunwu Qi1,
                            Rong Hou1 ✉, James R. Spotila1,2 & Zhihe Zhang1
                            Giant pandas (Ailuropoda melanoleuca) were historically hunted using dogs and are currently
                            threatened by free-roaming dogs and their associated diseases. To better understand the spatial
                            magnitude of this threat, we used a GIS approach to investigate edge effects of dogs on giant panda
                            habitat. We first examined two nature reserves with contrasting free-roaming dog populations:
                            Liziping, with many dogs (~0.44/km2), and Daxiangling, with few dogs (~0.14/km2). Spatial analysis
                            indicated that giant pandas at Liziping (but not Daxiangling) showed a shift in habitat use away from
                            populated areas consistent with a risk response to the foray distance of free-roaming dogs (10.9 km
                            path-distance). Most giant panda locations (86%) from the 2014 census in Liziping were clustered
                            around remote “dog-free zones.” Expanding this analysis across the entire giant panda range revealed
                            that 40% of panda habitat is within the foray distance of dogs. Our assessment will inform dog control
                            programs including monitoring, education, veterinary care, and other measures. We recommend
                            that reserves designated for the release of translocated pandas receive priority consideration for dog
                            control efforts. Only by understanding and managing complex interactions between humans, domestic
                            animals, and wild animals can we sustain natural systems in a world increasingly dominated by humans.

                            Nature reserves are primary sanctuaries for threatened species but are vulnerable to anthropogenic pressures
                            from outside their borders1,2. Edge effects, resulting in changes to species abundance and distribution, frequently
                            occur at the administrative borders of reserves3,4. Lower abundance near edges is indicative of negative edge
                            effects at the population level5 and mammalian carnivores are especially sensitive to this global threat3. Edge
                            effects are not confined to a few hundred meters, as was previously thought, but can persist 10 km from a habitat
                            edge, thus impacting the conservation of biodiversity at landscape scales6. However, this knowledge has failed to
                            translate into management reform for even some of our most iconic threatened species4,7. Due to a complexity of
                            socioecological factors, conservation efforts, such as reserve planning and corridor placement, have consistently
                            failed to incorporate mitigation techniques for addressing the role reserve edges play as population sinks for
                            species of concern.
                                Species interactions such as competition, disease transmission, and predation are altered at edges 8.
                            Collectively, these factors favor species that thrive in disturbed environments (such as invasive and generalist
                            species) at the expense of area sensitive native species9. The link between edge effects and invasive species has
                            been well documented10. Edges act as gateways for invasive species that can facilitate dispersal into the interior9.
                                One such invasive species is the domestic dog (Canis lupus familiaris). The global dog population is estimated
                            to exceed 700 million, making dogs the most abundant carnivore worldwide11. Approximately 75% of these dogs
                            are free-roaming12. When free-roaming dogs are associated with rural residences bordering nature reserves, con-
                            flicts with wildlife become inevitable13–15. Feral and free-roaming dogs have not received as much attention as
                            feral cats in the scientific literature16. However, dogs can have profound effects on wildlife. Dogs predate wild-
                            life13,17,18, spread disease19,20, harass wildlife21, cause impacted species to limit their habitat use22,23 and reduce
                            breeding success24. Dogs have even been implicated in the extinction of 11 species and are considered a potential
                            threat to over 188 additional species of concern25.
                                Predation, harassment and disease transmission by dogs can have large-scale edge effects in both fragmented
                            habitats26 and protected nature reserves13. Although landscape level edge effects due to feral and free-roaming
                            dogs have been reported for a handful of species15,23,27, there is an urgent need for basic guidelines outlining how
                            to address this problem. How do we incorporate the edge effects of dogs into the management of protected areas
                            to mitigate impacts on threatened species? Here, we provide strategies for how to measure potential impacts at a

                            1
                             Sichuan Key Laboratory of Conservation Biology for Endangered Wildlife, Chengdu Research Base of Giant Panda
                            Breeding, Chengdu, Sichuan, 610081, P.R. China. 2Drexel University, Philadelphia, PA, 19104, USA. 3Kilham Bear
                            Center, Lyme, NH, 03768, USA. ✉e-mail: hourong2000@panda.org.cn

Scientific Reports |   (2020) 10:10247 | https://doi.org/10.1038/s41598-020-66755-7                                                              1
Free-roaming dogs limit habitat use of giant pandas in nature reserves
www.nature.com/scientificreports/                                                                 www.nature.com/scientificreports

                            Source                               Maximum Distance
                            Meek, 199957                         30.0 km (Euclidean distance)
                            Sepúlveda, et al., 201539            4.3 km (Euclidean distance)
                            Scott & Causey, 197358               8.2 km (Euclidean distance)
                            Butler, Du Toit, & Bingham, 200413   6.0 km (Euclidean distance)
                            This study                           10.9 km (path-distance)

                            Table 1. Published maximum distances traveled by feral and free-roaming dogs.

                            landscape scale and outline how to incorporate feral-dog management into the landscape level planning of species
                            conservation using the giant panda as an example.
                                Giant pandas (Ailuropoda melanoleuca) require a minimum habitat size of 114 km2 to enable long-term per-
                            sistence of local populations28. Most nature reserves set aside for giant pandas exceed this area requirement.
                            However, nature reserves are often closely connected to human settlements. The Wolong Nature Reserve is the
                            most famous of the 67 panda reserves in China. It protects approximately 2,000 km2 of land but contains over
                            5,000 people in 1,436 households29. The human residents and their domestic animals directly and indirectly affect
                            giant pandas and the rate of habitat decline actually increased after the reserve was expanded in 197530. When
                            edge effects are accounted for, the suitable habitat within these reserves may fall well below the minimum area
                            requirement for giant pandas. Incorporating edge effects when determining effective habitat area (effective habitat
                            area = baseline habitat minus edge effects) is essential for the conservation and management of giant pandas. In
                            addition, edge effects should be considered when assessing suitable sites to release captive-reared or translocated
                            individuals. Many reserves with extant giant panda populations may already be at carrying capacity for the avail-
                            able effective habitat.
                                As with most bear species, giant pandas were historically hunted using dogs31 and are still harassed and chased
                            by dogs in nature (personal observation). Giant pandas are susceptible to mortality from canine distemper virus
                            and canine coronavirus32. Antibodies for both of these viruses were documented 25 years ago in free-roaming
                            dogs and giant pandas within the Wolong Nature Reserve in Sichuan, China33. A more recent study conducted
                            by our lab in Liziping Nature Reserve revealed that 21% of village dogs surveyed had positive antibody titers
                            for at least one of four viruses potentially lethal to giant pandas: canine distemper, parvovirus, rotavirus and/or
                            rabies34. In addition, 67% of village dogs were positive for gastrointestinal parasites, including two species known
                            to infect giant pandas: Ancylostoma caninum and Strongyloides sp.34. Despite these facts, resource managers do
                            not consider free-roaming dogs to be a major threat to wild giant pandas. There are no comprehensive plans in
                            place to remove free-roaming and feral dogs from nature reserves or to control dog populations in local villages
                            within the giant panda range.
                                Although considerably smaller than bears, dogs are effective hunters of many bear species. Dogs hunt in
                            packs and can move quickly through the dense vegetation used by giant pandas. Harassment by dogs can lead to
                            increased stress and energetically costly behavior for many species21. Even if dogs do not kill pandas directly, dog
                            attacks can result in wounds that lead to severe infections or cause stress related myopathies that ultimately result
                            in mortality. Dogs may pose an even greater threat to giant panda cubs, which are particularly vulnerable as the
                            most altricial of all bear species. Based on these risks, we predict that the presence of dogs will limit habitat use by
                            giant pandas and reduce the available effective habitat in nature reserves throughout the range of the giant panda.
                            Giant pandas within the “dog zone” will likely incur additional costs to reproduction and survival due to stress
                            and altered foraging behavior, making the “dog zone” either a sink habitat or an ecological trap.
                                Some previous models of giant panda habitat suitability have recognized that pandas avoid human-dominated
                            landscapes. These models incorporated distance from roads and residential areas in predicting habitat use28,35–38.
                            However, these models are potentially deficient where large populations of free-roaming dogs have access to giant
                            panda habitat. The edge effect of free-roaming dogs may be considerably greater than the distances used in these
                            studies.
                                Proximity to human residences is the most important factor in predicting rural dog distribution39. Foray
                            distances away from human residences differ considerably amongst movement studies of free-roaming and feral
                            dogs (Table 1) indicating a need for more comprehensive studies. Giant pandas (and wildlife in general) are
                            exposed to the threat of free-roaming dogs within this foray distance, even deep within nature reserves. In 2016
                            and 2017, we observed free-roaming dogs from local villages, alone and in packs of up to 4 individuals, traveling
                            in and out of the Liziping Nature Reserve34. These observations are consistent with previous studies of group
                            size of feral and free-roaming dogs in other ecosystems40. Group size of village dogs and a transition to more
                            “wolf-like” behavior are likely driven by factors such as the amount of provisioning provided by humans, the age,
                            gender and size of the dogs, and whether they routinely accompany their owners into wilderness areas41.
                                Given that nature reserves will vary in their susceptibility to free-roaming dog populations based on the local
                            density of human residences and their spatial relationship to the protected wildlife habitat, we sought to model the
                            potential edge effect of known populations of free-roaming dogs on giant panda habitat at a landscape and then
                            regional scale. In our study we asked the question: How can knowledge of the spatial extent of the free-roaming
                            dog threat be used to inform management for giant pandas at a regional scale? We first determined the distance
                            that a free-roaming dog will travel in giant panda habitat (foray distance) using a path-distance tool and direct
                            observations of dogs traveling away from villages into a local giant panda nature reserve. We then compared the
                            edge effect of free-roaming dogs on giant panda habitat use in two different nature reserves that differed in their
                            free-roaming dog populations. Once we had tested the utility of this approach, we expanded the scope of the
                            study to a regional scale and assessed the potential spatial threat of free-roaming dogs across all 67 giant panda

Scientific Reports |   (2020) 10:10247 | https://doi.org/10.1038/s41598-020-66755-7                                                               2
Free-roaming dogs limit habitat use of giant pandas in nature reserves
www.nature.com/scientificreports/                                                             www.nature.com/scientificreports

                            Figure 1. Study area map showing Liziping and Daxiangling Nature Reserves in Sichuan Province, China. Map
                            produced using ArcGIS 10.6.1: https://desktop.arcgis.com/en/.

                            nature reserves. Used sequentially, these steps provide a model approach for incorporating the spatial extent and
                            magnitude of edge effects of free-roaming dogs into species-oriented management plans at regional scales.

                            Materials and Methods
                            Study area. Liziping and Daxiangling Nature Reserves are important locations for giant panda conservation.
                            Both reserves are located in Sichuan Province, China, 102°-103°E, 28°-29°N (Fig. 1). These reserves cover 479
                            km2 and 295 km2, respectively; however, Daxiangling is adjacent to Long Canggou National Park, as well as the
                            Wawushan Nature Reserve, making the contiguous protected habitat area much larger. Liziping is the present
                            site for giant panda translocations and releases into the wild42. The landscape in both reserves is rugged, com-
                            prised of ridges and narrow valleys, although Daxiangling’s terrain is somewhat gentler. Elevation ranges from
                            1,330–4,800 m in Liziping and 1,300–3,400 m in Daxiangling. The average annual temperature is 14 °C in Liziping
                            and 16 °C in Daxiangling. The forests in both reserves are primarily broad-leaf evergreen at lower elevations
                            and transition into coniferous forest around 2,400 m. Dominant bamboo species in the understory at Liziping
                            are Yushania lineolata (at 2,000–2,600 m) and Arundinaria spanostachya (above 2,500 m). At Daxiangling,
                            Chimonobambusa szechuanensis is abundant below 2,400 m and Arundinaria faberi is found at 2,400–3,200 m.
                                Comprehensive surveys by our laboratory of the six village groups surrounding Liziping in 2017 found
                            that 212 of 334 owned dogs in those villages were free-roaming (64%). In contrast, a concurrent survey of a
                            large village group (including 11 villages) proximate to Daxiangling found that only 42 of 110 owned dogs were
                            free-roaming (38%). Thus, the threat of free-roaming dogs was far greater in Liziping than in Daxiangling.

                            Path-distance from residences. We digitized residences within 8 km of both nature reserves from aerial
                            imagery (2.5 m resolution Spot imagery from 2015 provided by ArcGIS Online) at a 1:10,000 scale. We classified
                            areas with housing density greater than 6.25 houses per km2 as “populated areas.” We then used these data and a
                            Digital Elevation Model (ASTER Global DEM, a product of METI and NASA) to run the path-distance tool in

Scientific Reports |   (2020) 10:10247 | https://doi.org/10.1038/s41598-020-66755-7                                                        3
Free-roaming dogs limit habitat use of giant pandas in nature reserves
www.nature.com/scientificreports/                                                               www.nature.com/scientificreports

                            ArcGIS. This created a continuous surface representing distance (combined vertical and horizontal) away from
                            populated areas. We incorporated vertical as well as horizontal distance because the terrain in our study area was
                            very steep and likely to influence access by free-roaming dogs. This process essentially created a series of buffers
                            that account for terrain complexity.

                            Dog-free zones. The maximum distance we observed for free-roaming dogs in Liziping, based on direct
                            field observations and preliminary results of an ongoing camera trap study, was 10.9 km (path-distance). This
                            distance is much greater than the 1.92 km distance estimated by previous studies as defining the impact of human
                            residences on giant panda habitat use36,43. Assuming that dogs can travel 10.9 km (path-distance) from their
                            residence, we extracted the area within this distance to map potentially “dog-free” zones for giant pandas within
                            both nature reserves. We assumed that all residences had the potential to house free-roaming dogs as the ability
                            to completely census each household was beyond the scope of this study.

                            Giant panda habitat and survey data.               Although over 23 studies have been published on giant panda
                            habitat selection44, there is currently no broadly accepted range-wide model of suitable giant panda habitat.
                            Therefore, for the purposes of this study, we used the IUCN designated extant giant panda range45 to represent
                            baseline available habitat for giant pandas within and surrounding each reserve. This dataset was readily available
                            for the entire range of the giant panda in China and allowed for repeatability. Henceforth, we refer to the IUCN
                            giant panda range dataset as “giant panda habitat” when referring to local spatial scales of habitat use and avail-
                            ability. We intentionally made the analysis for this study straightforward using readily available datasets so that
                            resource managers can easily repeat the process.
                                Location data for giant pandas were from the Fourth National Giant Panda Census46 in Sichuan Province. This
                            survey was carried out from 2011 to 2013. Giant panda occurrence locations were determined by the presence of
                            feces, footprints and foraging traces. We only had access to census data from the two reserves that we worked in,
                            Liziping and Daxiangling.

                            Giant panda habitat use.         To assess whether giant pandas were using habitat relative to its availability, we
                            first generated 100 random locations within the baseline giant panda habitat in each nature reserve (Liziping and
                            Daxiangling) to represent available habitat. We then compared the distribution of giant panda locations from the
                            giant panda survey data to the distribution of available habitat (as a function of path-distance from a residence).

                            Extrapolation to range-wide impacts. Using Liziping and Daxiangling Nature Reserves as case studies,
                            we extended our spatial analysis to incorporate the entire extant giant panda range (including all 67 giant panda
                            nature reserves). Given the vast extent of the giant panda range, we digitized populated areas at a scale of 1:40,000
                            within a buffer of 15 km (instead of at a scale of 1:10,000 as was done for the case studies). Again, we used Spot
                            imagery from 2015 provided by ArcGIS Online. The same steps were used to model path-distance away from
                            populated areas as were employed in the reserve level analysis described above. The edge effect of free-roaming
                            dogs (10.9 km path-distance) was then extracted from the extant giant panda range to represent effective giant
                            panda range.
                                We prioritized the 67 giant panda nature reserves for dog control programs by calculating the total area of
                            panda habitat within 10.9 km path-distance of human residences within each nature reserve. Nature reserves were
                            categorized as having a low dog threat if the total area of giant panda habitat within the “dog zone” was
Free-roaming dogs limit habitat use of giant pandas in nature reserves
www.nature.com/scientificreports/                                                             www.nature.com/scientificreports

                            Figure 2. Path-distance from populated areas in Liziping and Daxiangling Nature Reserves. “Populated areas”
                            are defined as areas with housing density greater than 6.25 houses per km2 (shown in black on the map). “Path
                            distance” is the combined vertical and horizontal distance (in meters) that a free-roaming dog would need
                            to travel away from a populated area to reach a certain point on the landscape. Red areas are within 3 km of a
                            populated area while green areas are 12–18 km from a populated area (based on path distance). Map produced
                            using ArcGIS 10.6.1: https://desktop.arcgis.com/en/.

                            Figure 3. Path-distance from populated areas within the human impact zone (1,920 m shown in dark gray)
                            and the maximum foray distance (10,900 m shown in light gray) of free-roaming dogs in Liziping (with panda
                            locations from 2014 Fourth National Giant Panda Census). White areas indicate “dog-free zones”. However,
                            there were no dogs reported in Daxiangling Nature Reserve. Map produced using ArcGIS 10.6.1: https://
                            desktop.arcgis.com/en/.

                            required for long-term population viability. In contrast, the effective giant panda habitat for Daxiangling (271.1
                            km2), modeled based on the effect of humans without free-roaming dogs, was not reduced by edge effects as the
                            villages were more than 2 km from the reserve boundary. However, if the dog population grows at Daxiangling,
                            and dogs enter the reserve, we predict that 76% of the current habitat will be affected (Table 2).

                            Use versus availability. By comparing habitat use to habitat availability, we found that at Liziping the habi-
                            tat use of giant pandas (as a function of path-distance from populated areas) was shifted approximately 7 km far-
                            ther away from populated areas than would be expected given the distribution of available panda habitat (Fig. 5).
                            This shift in distance is two standard deviations farther away than the mean distance to available protected panda
                            habitat (5.1 km vs. 12.4 km). Approximately 83% of the giant panda locations from the 2014 Fourth National
                            Giant Panda Census at Liziping were greater than 10.9 km from a residence, much farther away than would be
                            predicted based on the available habitat within the reserve. This pattern was not seen in Daxiangling, where few
                            dogs lived in the local villages and even fewer were free-roaming. The distribution of giant panda locations in

Scientific Reports |   (2020) 10:10247 | https://doi.org/10.1038/s41598-020-66755-7                                                          5
Free-roaming dogs limit habitat use of giant pandas in nature reserves
www.nature.com/scientificreports/                                                                     www.nature.com/scientificreports

                            Figure 4. Baseline and effective giant panda habitat for Liziping and Daxiangling Nature Reserves. (a) Baseline
                            protected and unprotected giant panda habitat in and surrounding Liziping. (b) Effective giant panda habitat
                            at Liziping when the influence of free-roaming dogs is incorporated (with panda locations from 2014 Fourth
                            National Giant Panda Census). (c) Baseline and effective giant panda habitat in and surrounding Daxiangling.
                            d) Effective giant panda habitat at Daxiangling (with panda locations from 2014 Fourth National Giant Panda
                            Census). Map produced using ArcGIS 10.6.1: https://desktop.arcgis.com/en/.

                                                                        Effective Giant Panda   Effective Giant Panda
                                          Baseline Giant Panda          Habitat I (human        Habitat II (human and
                                          Habitat within each reserve   impact)                 dog impact)
                            Liziping      385.2 km2                     380.3 km2               108.7 km2
                            Daxiangling   271.2 km2                     271.1 km2               63.9 km2*

                            Table 2. Difference between baseline protected area (within the reserve) and effective protected area of giant
                            panda habitat for two nature reserves. Areas provided were based on the impact of humans alone (scenario I:
                            within the reserve and greater than 1.9 km path-distance from a populated area) and human and free-roaming
                            dogs together (scenario II: within the reserve and greater than 10.9 km path-distance from a populated area).
                            *Although surveys indicate no current dog problem in Daxiangling, this is the effective area that would result
                            from a future free-roaming dog population in Daxiangling.

                            Daxiangling did show a small shift away from residences (6.0 km vs. 7.8 km) consistent with the 1.92 km edge
                            effect distance (due to humans alone) suggested by previous authors35,36. This shift was within one standard devi-
                            ation of the mean of the available habitat.

                            Extrapolation to range-wide impacts.             Range-wide analysis of potential edge effects of free-roaming
                            dogs showed that 40% of the total giant panda range was within the foray distance of free-roaming dogs (Table 3).

Scientific Reports |   (2020) 10:10247 | https://doi.org/10.1038/s41598-020-66755-7                                                          6
www.nature.com/scientificreports/                                                                        www.nature.com/scientificreports

                            Figure 5. Histogram showing available habitat versus habitat used by giant pandas as a function of path-
                            distance from a populated area in two nature reserves in Sichuan Province, China. (a) Liziping (many dogs)
                            and (b) Daxiangling (few dogs). Means and standard deviations represent the average path-distance from a
                            populated area for available habitat (shown in red bars and based on 100 random locations within suitable
                            habitat in each reserve) and used habitat (shown in blue bars and based on 46 and 50 giant panda locations from
                            the 2014 Fourth National Giant Panda Census).

                                                                   Total area       Edge effect area   Dog free area   Percent dog free

                            Extant giant panda range               25,710 km2       10,240 km2         15,470 km2      60%

                            Extant giant panda range in reserves   13,720 km    2
                                                                                    4,060 km   2
                                                                                                       9,660 km   2
                                                                                                                       70%

                            Table 3. Extent of potential edge effects of dogs on giant panda range inside and outside all 67 protected nature
                            reserves.

                            Giant panda range within reserves scored better with only 30% of habitat occurring within the foray distance of
                            free-roaming dogs. However, edge effect was not distributed evenly across the geographic range of giant pandas
                            (Fig. 6). The majority of edge effect was concentrated in the southern portion of the giant panda range with 6,730
                            km2 (66%) of the edge affected area south of the 31st parallel. Gentler topography along the Dadu and Qingyi
                            rivers in the south facilitated development and provided shorter path-distances to giant panda habitat. In the
                            northern portion of the range, dramatic topographic relief, as well as lower human population density, reduced
                            the vulnerability of giant panda populations to the threat of free-roaming dogs.
                                The amount of habitat within the dog zone in each reserve ranged from none (in several remote reserves) to
                            367 km2 (in Wolong Nature Reserve). Our classification of the giant panda nature reserves based on dog threat
                            categories resulted in five nature reserves falling into the high risk category (including Liziping and Wolong
                            Nature Reserves). Twenty-two of the reserves were classified as moderate while the majority of reserves (40 in
                            total) received a low threat classification (Fig. 7).

                            Discussion
                            Mapping the potential spatial extent of the edge effect of free-roaming dogs in giant panda habitat proved very
                            informative. Without considering the threat of free-roaming dogs, Liziping Nature Reserve potentially pro-
                            vided more protected habitat for giant pandas than did Daxiangling Nature Reserve. However, Daxiangling had
                            substantially fewer free-roaming dogs in villages and none observed within the reserve. In contrast, Liziping
                            had hundreds of free-roaming dogs with easy access to the habitat within the reserve. The spatial pattern of
                            giant panda distribution in Liziping was indicative of a risk effect (direct lethal, indirect lethal, and/or indirect
                            non-lethal) in response to free-roaming dogs. The threat posed by free-roaming dogs has the potential to reduce

Scientific Reports |   (2020) 10:10247 | https://doi.org/10.1038/s41598-020-66755-7                                                            7
www.nature.com/scientificreports/                                                              www.nature.com/scientificreports

                            Figure 6. Range-wide assessment of potential edge effects of free-roaming dogs on giant pandas. Gray areas
                            show extant giant panda range that is within the travel distance of free-roaming dogs (10.9 km path-distance).
                            Green areas show potentially “dog free” giant panda range. Map produced using ArcGIS 10.6.1: https://desktop.
                            arcgis.com/en/.

                            the effective giant panda habitat by 72% at Liziping. In contrast, this was not the case in Daxiangling, where the
                            spatial distribution of giant panda locations appeared to be in response to human impact alone (without dogs).
                            Thus, we estimated that the available effective giant panda habitat at Daxiangling was 271 km2, much greater
                            than the effective giant panda habitat in Liziping (109 km2). The larger total area occupied by the Liziping Nature
                            Reserve (479 km2) was deceptive since the effective habitat area may be considerably smaller (109 km2).
                                The spatial clumping of giant pandas within the “dog free zone” suggests an edge effect of free-roaming dogs
                            on habitat use by giant pandas in Liziping Nature Reserve. However, there are other potential explanations for
                            this spatial pattern. For example, it is possible that the most highly suitable habitat for giant pandas is coinci-
                            dentally found within the “dog-free zone.” There are also many additional threats to giant pandas originating
                            from populated areas. Incidental take from poaching, bamboo harvesting, collection of medicinal plants and
                            fuelwood collection have all been implicated as human factors influencing giant panda populations and hab-
                            itat use35,47. According to Liu et al.35, however, effects from these disturbances are limited to a 1.92 km buffer
                            around populated areas whereas our study suggests that the effect of free-roaming dogs extends to ~10.9 km
                            path distance away from populated areas. Livestock use of giant panda habitat at Wanglang Nature Reserve was
                            shown to reduce giant panda habitat by 34% but the distance from populated areas was not a factor measured in
                            this study48. Impacts from livestock likely extend farther than the 1.92 km buffer distance cited for other human
                            factors by Liu et al.35. The effect of livestock may be compensatory or additive when combined with the effect of
                            free-roaming dogs.
                                A recent study estimated a carrying capacity of 163 individuals based on bamboo density (Bashania spanos-
                            tachya) and high quality habitat in Liziping42. However, the existing giant panda population is a small fraction
                            of this estimate, possibly due to the effective habitat area being much smaller than the suitable baseline habitat.

Scientific Reports |   (2020) 10:10247 | https://doi.org/10.1038/s41598-020-66755-7                                                          8
www.nature.com/scientificreports/                                                                 www.nature.com/scientificreports

                            Figure 7. Giant panda nature reserves prioritized for dog control programs based on the area of panda habitat
                            within 10.9 km path distance of human residences (Low < 50 km2, Moderate  165 km2).
                            Map produced using ArcGIS 10.6.1: https://desktop.arcgis.com/en/.

                            If the issue of free-roaming dogs is addressed at Liziping, the effective giant panda habitat area should increase,
                            simultaneously increasing the actual carrying capacity of the reserve.
                                 The edge effect of dogs on giant panda habitat use is likely compounded by the spread of disease. In a recent
                            study in our laboratory, 5% of surveyed dogs surrounding Liziping Nature Reserve tested positive for exposure to
                            canine distemper virus and 13% tested positive for exposure to rabies virus34, both of which are potentially lethal
                            for giant pandas. The importance of vaccinating and monitoring village dogs for canine distemper virus, canine
                            coronavirus and rabies cannot be overstated. Dogs travel great distances (up to 35 km) when rabid49. At this effect
                            distance, an outbreak of rabies would expose all individual giant pandas in a reserve to the disease threat.
                                 Future studies should focus on the factors that influence dog foray distance. Does group size determine foray
                            distance and influence “wolf-like” behaviors such as harassing and preying on wildlife? What role does size, age
                            and gender of individual dogs play? Clearly, dog owners influence the likelihood of dogs entering nature reserves
                            through variability in restriction methods. Owners may also unintentionally encourage forays by initially taking
                            their dogs on resource extraction excursions into the reserve and/or by limiting the amount of supplemental food
                            the dogs receive. We theorize that dog foray distance is a probabilistic variable that would best be modeled using
                            a known distribution once enough spatial data is collected to set the parameters. Generating a surface of known
                            probability of dog encounters would provide a more mechanistic understanding of this relationship and more
                            clearly inform management efforts. For example, at what threshold of dog impact do giant pandas alter their hab-
                            itat use? Do they respond to even low risks of encounters? Anecdotal evidence suggests that even solitary medium
                            sized dogs will harass giant pandas but it is likely that only groups of dogs can inflict serious injuries. Solitary dogs
                            still pose lethal risks to giant panda cubs and can introduce pathogens.
                                 Based on our case studies in Liziping and Daxiangling, free-roaming dogs have the potential to limit giant
                            panda habitat use, reproductive success and population persistence. To prioritize future studies and management

Scientific Reports |   (2020) 10:10247 | https://doi.org/10.1038/s41598-020-66755-7                                                                9
www.nature.com/scientificreports/                                                                www.nature.com/scientificreports

                            efforts, we provided spatial data identifying areas potentially threatened by free-roaming dogs within the entire
                            range of giant pandas (Fig. 6). Our basic spatial analysis is only a preliminary step. The resulting maps can be
                            used as a tool for managers to select the most crucial nature reserves (shown in red in Fig. 7) for more extensive
                            surveys of the local free-roaming dog populations. Where the distribution of free-roaming dogs and giant pandas
                            overlap, we recommend the initiation of extensive GPS telemetry studies, as was recently performed in Foping
                            Nature Reserve51. In Foping Nature Reserve (33.658°N 107.807°E), researchers tracked 85 dogs from two villages
                            near the reserve, of which 41 traveled outside of the village in 8 packs51. This study did not indicate the total num-
                            ber of dogs nor the proportion that were free roaming in villages surrounding the reserve. However, given that
                            nearly 50% of the monitored dogs left the village to enter giant panda habitat, it is clear that the dog problem in
                            Liziping is not an isolated issue and the preliminary spatial analysis we present here is of value to other reserves
                            in the network.
                                The study in Foping51 provides a template for future in-depth studies of free-roaming dog impacts on giant
                            panda habitat use. We found that the majority of edge affected giant panda habitat was south of 31°N latitude and
                            suggest that future studies be initiated in the southern mountain ranges which support the smallest and most
                            isolated populations of giant pandas28. Giant pandas struggling to survive in this fragmented habitat are already
                            susceptible to the suite of demographic and genetic threats facing small populations. Controlling free-roaming
                            dog populations surrounding these southern reserves is a critical, but currently unutilized, management tool for
                            preventing these small giant panda populations from going locally extinct.
                                As part of The National Conservation and Management Plan for the Giant Panda and its Habitat, the reintro-
                            duction of giant pandas has been initiated in two nature reserves (Wolong and Liziping). When selecting sites for
                            translocations and reintroductions of giant pandas, there are many factors to consider: habitat availability and
                            quality, the size and sex ratio of the existing population, bamboo species composition and abundance, connectiv-
                            ity to other giant panda populations, threats from competitors, disease and predators, road density and proximity
                            to populated areas. The threats posed by local free-roaming dog populations must be added to the list of factors
                            addressed for any comprehensive assessment aimed at selecting sites for conservation actions (particularly the
                            release of naïve translocated individuals). Thus, these reserves should also receive top priority for camera traps
                            and field surveys for dogs. Results indicating an effect of dogs on giant panda habitat use should trigger dog con-
                            trol programs focused on education for residents, free neuter and vaccination clinics and removal of feral dogs.
                                Where they are abundant and in close proximity to giant panda habitat, dogs are likely a greater threat to
                            giant pandas than snares, bamboo harvesting, and natural predators. As such, free-roaming and feral dogs must
                            be removed from nature reserves. Spay and neuter programs, as well as vaccination programs, should be imple-
                            mented in all villages within 10 km of reserves. Comprehensive approaches that involve dog licensing and col-
                            laring need to be organized by local governments and implemented by village leaders. Local villagers are very
                            supportive of giant panda conservation but do not realize that their dogs pose serious threats to giant pandas. In
                            addition, the occupation and lifestyle of the residents influence the number of dogs per household and the level of
                            control exerted over those dogs52. Education programs for local villagers that address the threat of free-roaming
                            dogs are essential. Research shows that responsible management can greatly reduce the ecological impact of dogs
                            in protected areas50.
                                Working with local governments can facilitate the implementation of novel solutions based on local needs.
                            The employment of local villagers to help monitor the feral and free-roaming dog populations will improve the
                            success of these programs. Consideration for the ethical treatment of dogs removed from reserves should be
                            incorporated into all stages of the dog management plan through consultation with the Society for the Prevention
                            of Cruelty to Animals or similar local groups. The dog population in China was estimated at 250 million in 2012
                            based on the human census and assumed human: dog ratios11. The threat these animals pose to giant pandas, the
                            national treasure of China, should serve to galvanize control efforts. The creation of the Giant Panda National
                            Park provides an opportunity to standardize and enforce dog control programs throughout the range of the giant
                            panda.
                                Setting aside nature reserves for giant pandas has protected thousands of other species53. Addressing the dog
                            problem in these reserves will also benefit biodiversity by protecting other vertebrate species from dog predation,
                            harassment and disease transmission. Areas with feral dog populations have lower species diversity of terrestrial
                            mammals and lower abundance of extant species54. The mountains of Sichuan are considered one of the world’s
                            top biodiversity hotspots55. It is likely that free-roaming and feral dogs are a major unrecognized threat to the
                            biodiversity of the region. In addition, China has the second highest incidence of human rabies cases56; therefore,
                            controlling the dog population will also benefit public health for rural residents.
                                Our study illustrates that the effectiveness of nature reserves depends upon our understanding and manage-
                            ment of complex interactions between the still increasing populations of humans and their domestic animals and
                            the wild animals such as giant pandas that we are trying to protect. While widely acknowledged by ecologists,
                            edge effects at broad spatial scales are rarely addressed explicitly when tackling the numerous threats facing spe-
                            cies of concern. Recent studies of the edge effects of intrusive free-roaming dog populations on the distribution
                            of threatened species in nature reserves23,27,51 indicate a need for comprehensive changes to policy for managing
                            biodiversity in protected areas. Only through the development and application of this knowledge can we hope to
                            sustain natural systems in a world increasingly crowded by humans.

Scientific Reports |   (2020) 10:10247 | https://doi.org/10.1038/s41598-020-66755-7                                                           10
www.nature.com/scientificreports/                                                                             www.nature.com/scientificreports

                            Data availability
                            The locations of human residences surrounding Liziping and Daxiangling Nature Reserves as well as populated
                            areas for the entire giant panda range will be made available on http://databasin.org/datasets/ upon acceptance
                            of the manuscript. The 2014 National Giant Panda Census data are sensitive, protected by the State Forestry
                            Administration, P.R.C., and are not generally available. We only had access to the giant panda census data for the
                            two reserves that we studied in detail (Liziping and Daxiangling).

                            Received: 12 November 2019; Accepted: 7 May 2020;
                            Published: xx xx xxxx

                            References
                             1. Laurance, W. F. et al. Averting biodiversity collapse in tropical forest protected areas. Nature 489, 290–294 (2012).
                             2. Lovejoy, T. E. Protected areas: A prism for a changing world. Trends Ecol. Evol. 21, 329–333 (2006).
                             3. Woodroffe, R. & Ginsberg, J. R. Edge effects and the extinction of populations inside protected areas. Science (80-.). 280, 2126–2128
                                (1998).
                             4. Revilla, E., Palomero, F. & Delibes, M. Edge-core effects and the effectiveness of traditional reserve in conservation: Eurasian badgers
                                in Dona˜na National Park. Conserv. Biol. 15, 148–158 (2001).
                             5. Laurance, S. G., Stouffer, P. & Laurance, W. Effects of road clearings on movement patterns of understory rainforest birds in Central
                                Amazonia. Conserv. Biol. 18, 1099–1109 (2004).
                             6. Laurance, W. F. Do edge effects occur over large spatial scales? Trends Ecol. Evol. 15, 134–135 (2000).
                             7. Brodie, J. F., Giordano, A. J. & Ambu, L. Differential responses of large mammals to logging and edge effects. Mamm. Biol. 80, 7–13
                                (2015).
                             8. Fagan, W. F., Cantrell, R. S. & Cosner, C. How habitat edges change species interactions. Am. Nat. 153, 165–182 (1999).
                             9. Paton, W. C. The effect of edge on avian nest success: how strong is the evidence? Conserv. Biol. 8, 17–26 (1994).
                            10. Didham, R. K., Tylianakis, J. M., Gemmell, N. J., Tatyana, R. A. & Ewers, R. M. Interactive effects of habitat modification and species
                                invasion on native species decline. Trends Ecol. Evol. 22, 489–496 (2007).
                            11. Hughes, J. & Macdonald, D. W. A review of the interactions between free-roaming domestic dogs and wildlife. Biol. Conserv. 157,
                                341–351 (2013).
                            12. Massei, G., Miller, L. A. & Killian, G. Immunocontraception to control rabies in dog populations The Soap Box in dog populations.
                                USDA Natl. Wildl. Res. Cent. - Staff Publ. 1275, (2010).
                            13. Butler, J. R. A., Du Toit, J. T. & Bingham, J. Free-ranging domestic dogs (Canis familiaris) as predators and prey in rural Zimbabwe:
                                Threats of competition and disease to large wild carnivores. Biol. Conserv. 115, 369–378 (2004).
                            14. Srbek-Araujo, A. C. & Chiarello, A. G. Domestic dogs in Atlantic forest preserves of south-eastern Brazil: a camera-trapping study
                                on patterns of entrance and site occupancy rates. Brazilian. J. Biol. 68, 771–779 (2008).
                            15. Vanak, A. T. & Gompper, M. E. Dogs Canis familiaris as carnivores: Their role and function in intraguild competition. Mamm. Rev.
                                39, 265–283 (2009).
                            16. Young, J. K., Olson, K. A., Reading, R. P., Amgalanbaatar, S. & Berger, J. Is Wildlife Going to the Dogs? Impacts of Feral and Free-
                                roaming Dogs on Wildlife Populations. Bioscience, https://doi.org/10.1525/bio.2011.61.2.7 (2011).
                            17. Pierce, R. J. & Sporle, W. Causes of kiwi mortality in Northland. Conserv. Advis. Sci. Notes 1–6 (1997).
                            18. Wierzbowska, I. A., Hedrzak, M., Popczyk, B., Okarma, H. & Crooks, K. R. Predation of wildlife by free-ranging domestic dogs in
                                Polish hunting grounds and potential competition with the grey wolf. Biol. Conserv. 201, 1–9 (2016).
                            19. Woodroffe, R. & Ginsberg, J. R. Conserving the African wild dog Lycaon pictus. I. Diagnosing and treating causes of decline. Oryx
                                33, 132–142 (1999).
                            20. Mamaev, L. V. et al. Characterisation of morbilliviruses isolated from Lake Baikal seals (Phoca sibirica). Vet. Microbiol. 44, 251–259
                                (1995).
                            21. Silva-Rodríguez, E. A. & Sieving, K. E. Influence of Care of Domestic Carnivores on Their Predation on Vertebrates. Conserv. Biol.
                                25, 808–815 (2011).
                            22. Lenth, B. E., Knight, R. L. & Brennan, M. E. The effects of dogs on wildlife communities. Nat. Areas J. 28, 218–227 (2008).
                            23. Silva-rodríguez, E. A. & Sieving, K. E. Domestic dogs shape the landscape-scale distribution of a threatened forest ungulate. 150,
                                103–110 (2012).
                            24. Gingold, G. & Geffen, E. Effect of guard dogs on the behavior and reproduction of gazelles in cattle enclosures on the Golan Heights.
                                12, 155–162 (2009).
                            25. Doherty, T. S. et al. The global impacts of domestic dogs on threatened vertebrates. Biol. Conserv. 210, 56–59 (2017).
                            26. Torres, P. C. & Prado, P. I. Domestic dogs in a fragmented landscape in the Brazilian Atlantic Forest: abundance, habitat use and
                                caring by owners. Brazilian. J. Biol. 70, 987–994 (2010).
                            27. Lacerda, A. C. R., Marinho-Filho, W. M. & Tomas, J. Domestic dogs as an edge effect in the Bras´ılia National Park, Brazil:
                                interactions with native mammals. Anim. Conserv. 5, 477–487 (2009).
                            28. Qing, J. et al. The minimum area requirements (MAR) for giant panda: An empirical study. Sci. Rep. 6, 1–9 (2016).
                            29. Hull, V. et al. Peek into a home for pandas and people. in Pandas and People, Coupling Human and Natural Systems for Sustainability
                                (ed. Liu, Junguo, Hull, V., Yang, W., Vina, A., Che, X., Ouyang, Z. & Zhang, H.) 37–48 (New York: Oxford University Press., 2016).
                            30. Liu, J. et al. Ecological Degradation in Protected Areas: The Case of Wolong Nature Reserve for Giant Pandas. 292, 98–101 (2001).
                            31. Sheldon, W. G. The wilderness home of the giant panda. (University of Massachusetts Press, 1975).
                            32. Qiu, X. & Mainka, S. Review of mortality of the giant panda (Ailuropoda melanoleuca). J. Zoo Wildl. Med. 24, 425–429 (1993).
                            33. Mainka, S. A., Qiu, X., He, T. & Appel, M. J. Serologic survey of giant pandas (Ailuropoda melanoleuca), and domestic dogs and cats
                                in the Wolong Reserve, China. J. Wildl. Dis. 30, 86–89 (1994).
                            34. Yan, X. et al. Dogs and Disease Threats to Giant Pandas in China. J. Wildl. Manage. 84, 268–276 (2020).
                            35. Liu, J., Ouyang, Z. & Taylor, W. W. A framework for evaluating the effects of human factors on wildlife habitat: the case of giant
                                pandas. Conserv. Biol. 13, 1360–1370 (1999).
                            36. Xu, W. et al. Designing a conservation plan for protecting the habitat for giant pandas in the Qionglai mountain range, China. Divers.
                                Distrib. 12, 610–619 (2006).
                            37. Qi, D. et al. Using habitat models to evaluate protected area designing for giant pandas. Folia Zool. 64, 56–64 (2015).
                            38. Zhao, C. et al. Relationship between human disturbance and endangered giant panda (Ailuropoda melanoleuca) habitat use in the
                                Daxiangling Mountains. Oryx 51, 146–152 (2017).
                            39. Sepúlveda, M., Pelican, K., Cross, P., Eguren, A. & Singer, R. Fine-scale movements of rural free-ranging dogs in conservation areas
                                in the temperate rainforest of the coastal range of southern Chile. Mamm. Biol. 80, 290–297 (2015).
                            40. L. Boitani & Ciucii, P. Comparative social ecology of feral dogs and wolves. 49–72 (1995).
                            41. Vanak, A. B. I. T. & Gompper, M. E. Dogs Canis familiaris as carnivores: their role and function in intraguild competition. 39,
                                265–283 (2009).

Scientific Reports |   (2020) 10:10247 | https://doi.org/10.1038/s41598-020-66755-7                                                                                 11
www.nature.com/scientificreports/                                                                             www.nature.com/scientificreports

                            42. Yang, Z. et al. Reintroduction of the giant panda into the wild: A good start suggests a bright future. Biol. Conserv. 217, 181–186
                                (2018).
                            43. Liu, J. et al. A framework for evaluating the effects of human factors on wildlife habitat: The case of giant pandas. Conserv. Biol. 13,
                                1360–1370 (1999).
                            44. Hull, V. et al. A synthesis of giant panda habitat selection. Ursus 25, 148–162 (2014).
                            45. Swaisgood, R., Wang, D. & Wei, F. Ailuropoda melanoleuca (errata version published in 2017). The IUCN Red List of Threatened
                                Species 2016 (2016).
                            46. State Forestry Administration, P. R. C. Report of the Fourth National Giant Panda Census. (Science Publishing House, Beijing,
                                China., 2015).
                            47. Yiming, L., Zhongwei, G., Qisen, Y., Yushan, W. & Niemelä, J. The implications of poaching for giant panda conservation. Biol.
                                Conserv. 111, 125–136 (2003).
                            48. Li, B. V., Pimm, S. L., Li, S., Zhao, L. & Luo, C. Free-ranging livestock threaten the long-term survival of giant pandas. Biol. Conserv.
                                216, 18–25 (2017).
                            49. Gompper, M. E. Free-ranging dogs and wildlife conservation. (Oxford University Press, 2013).
                            50. Parsons, A. W. et al. The ecological impact of humans and dogs on wildlife in protected areas in eastern North America. Biol.
                                Conserv. 203, 75–88 (2016).
                            51. Jin, Y. et al. Canine distemper viral infection threatens the giant panda population in China. Oncotarget 8, 113910–113919 (2017).
                            52. Hsu, Y., Liu Severinghaus, L. & Serpell, J. A. Dog keeping in Taiwan: its contribution to the problem of free-roaming dogs. J. Appl.
                                Anim. Welf. Sci. 6, 1–23 (2003).
                            53. Li, B. V. & Pimm, S. L. China’s endemic vertebrates sheltering under the protective umbrella of the giant panda. Conserv. Biol. 30,
                                329–339 (2016).
                            54. Zapata-Ríos, G. & Branch, L. C. Altered activity patterns and reduced abundance of native mammals in sites with feral dogs in the
                                high Andes. Biol. Conserv. 193, 9–16 (2016).
                            55. Mittermeier, R. A., Turner, W. R., Larsen, F. W., Brooks, T. M. & Gascon, C. Global Biodiversity Conservation: The Critical Role of
                                Hotspots. in Biodiversity Hotspots (eds. Habel, F. E. & J.C., Z.) 3–8 (Springer-Verlag Berlin Heidelberg, 2011).
                            56. Knobel, D. L. et al. Re-evaluating the burden of rabies in Africa and Asia. Bull. World Health Organ. 83, 360–368 (2005).
                            57. Meek, P. D. The movement, roaming behaviour and home range of free-roaming domestic dogs. Canis lupus familiaris, in coastal
                                New South Wales. Wildl. Res. 26, 847 (1999).
                            58. Scott, M. D. & Causey, K. Ecology of feral dogs in Alabama. J. Wildl. Manage. 37, 253–265 (1973).

                            Acknowledgements
                            This research was funded by the Chengdu Research Base of Giant Panda Breeding and Drexel University. Final
                            line edits were contributed by Sophia Courtney and Teresa Valentine. We acknowledge support from the State
                            Forestry Administration, Sichuan Forestry Bureau, Liziping and Daxiangling National Nature Reserves.

                            Author contributions
                            J.R.S. and B.K. conceived of the research. R.C. collected data, performed the spatial analysis and led the writing,
                            with assistance with analysis and editorial input from J.R.O., J.R.S., D.Q., R.H., B.K. and Z.Z. X.Y. and W.B.
                            collected data and contributed to analysis and writing.

                            Competing interests
                            The authors declare no competing interests.

                            Additional information
                            Correspondence and requests for materials should be addressed to R.H.
                            Reprints and permissions information is available at www.nature.com/reprints.
                            Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
                            institutional affiliations.
                                          Open Access This article is licensed under a Creative Commons Attribution 4.0 International
                                          License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
                            format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre-
                            ative Commons license, and indicate if changes were made. The images or other third party material in this
                            article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the
                            material. If material is not included in the article’s Creative Commons license and your intended use is not per-
                            mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
                            copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

                            © The Author(s) 2020

Scientific Reports |   (2020) 10:10247 | https://doi.org/10.1038/s41598-020-66755-7                                                                                  12
You can also read